SPECT/CT stabilizes the interpretation of somatostatin receptor scintigraphy findings: a retrospective analysis of inter-rater agreement

Abstract

Correlating the anatomical information from CT with the functional information from SPECT improves diagnostic accuracy of somatostatin-receptor-scintigraphy (SRS) in patients with neuroendocrine tumors (NET). The aim of the present study was to investigate the impact of dual modality SPECT/CT on the inter-rater agreement in SRS. Twenty-five unselected patients with suspected or histologically proven NET in whom whole body planar imaging and low-dose SPECT/CT had been performed after injection of 200 MBq In-111-octreotide were included retrospectively. Images were interpreted independently by 2 nuclear medicine physicians, an experienced one and an inexperienced one. Both readers first re-evaluated the planar whole-body images alone, then added the SPECT images, and finally the CT-images. Lesions with pathologically increased tracer uptake were categorized according to the following 3-point score: equivocal, probably pathologic, and definitely pathologic. Cohen's linear-weighted kappa coefficient kappa was used to quantify inter-rater agreement. A total number of 50 lesions were described in 23 of the 25 patients. The two readers showed only moderate agreement in the interpretation of the planar findings (kappa = 0.593). Agreement improved to substantial by adding SPECT (kappa = 0.736) and to very good by adding SPECT/CT (kappa = 0.860). SPECT/CT resulted in up-staging of 18% of the lesions and down-staging of 12% compared to planar + SPECT (experienced reader). In addition, SPECT/CT tended to reduce the frequency of indefinite scores (equivocal, probably pathologic), from 18% in planar + SPECT to 6% (p = 0.065). Change of lesion localization by SPECT/CT tended to contribute to the change of lesion score (p = 0.055). The present results suggest that low-dose SPECT/CT stabilizes report quality in SRS by improving inter-rater agreement.